Content uploaded by Federico Milano
Author content
All content in this area was uploaded by Federico Milano on Jan 19, 2022
Content may be subject to copyright.
1
Co-ordinated Grid Forming Control of AC-side-connected
Energy Storage Systems for Converter-Interfaced
Generation
Junru Chen1*, Muyang Liu1, Renqi Guo2, Nan Zhao2, Federico Milano2, and Terence O’Donnell2
Xinjiang University, Urumqi, China
University College Dublin, Dublin, Ireland
Abstract
Grid forming control of converter interfaced generation (CIG) requires some form
of energy storage to be coupled with the generation. Energy storage systems (ESSs)
can be coupled to the CIG either on the DC or the AC side of the power converter.
When placed on the DC side, the ESS can provide damping of the variability in the
generation but would require significant modification to the wind turbine
hardware. The solution with an ESS connected to the AC side is simpler to
implement with existing wind turbines but fails to provide damping of the CIG
generation. This paper proposes a grid forming control strategy, based on virtual
synchronous generator (VSG) control, which allows the ESS installed at the AC-side
of the converter to have the same features and dynamic behaviour as those obtained
from placement on the DC-side of the converter. In addition, the proposed control
can also limit the exchanged power of the ESS within its rating for a safe operation.
The proposed control is validated via a detailed Electro-Magnetic Transient (EMT)
model and its impact on the grid is quantified via the case study of the All-Island
Irish transmission system. Simulation results show that only a small ESS capacity
can ensure that the frequency variance satisfies the grid code requirement even in
the situation of a very high CIG penetration.
Keywords—Energy Storage System, Grid forming Control, Converter-Interfaced
Generation, Damping, High Wind Penetration.
*Corresponding author, email: junru.chen.1@ucdconnect.ie
2
1 Introduction
1.1 Background
Electrical power systems are in the transition towards a 100% penetration of
converter-interfaced generation (CIG), such as from solar panels and wind turbines.
The replacement of traditional synchronous generators (SGs) by the CIGs reduces
system inertia, thereby potentially leading to lower frequency support and potential
frequency instability issues following a contingency or a sudden power imbalance in
the grid [1]. In order to maintain the stability of the system under a high penetration of
CIGs, it is desirable that at least some of the grid-tied converters have grid forming
capability and provide either an emulated inertial response or a fast frequency
regulation [2].
1.2 Literature review
The operation of CIG as grid forming converters implies the existence of an energy
store which can be utilized to address supply-demand imbalance during transients,
generally either in the form of an emulated inertia or droop response. Several
possibilities exist for the provision of the energy store. De-loaded operation of the CIG
is a method used to reserve a certain amount of the available power as “stored kinetic
energy” by purposely setting the CIG operating point retarded or advanced from its
optimum [3]. However, the disadvantage of this method is that it leads to an under-
utilization of the renewable energy source. This has economic implications for the
system and the CIG [4], since the renewable energy has the lowest marginal cost.
Alternatively, virtual inertia control strategies in wind turbines can link the speed of
the turbine to the grid frequency to extract the power from the energy stored in the
rotating mass of the turbine. However, the kinetic energy released from the turbine
reduces the rotor speed consequently resulting in generation reduction [5] and wind
3
energy waste for a long-term operation [6].
Alternatively in CIG, the transient power required during a disturbance can be
supplied from the DC-link capacitance. In fact, the voltage variation on the DC-link
capacitor of the grid-tied converter, has been shown to be equivalent to the inertial
response of an SG [7] and a virtual inertia control has been proposed firstly in [8] and
then improved in [9] and [10], which uses this DC voltage dynamic to emulate inertia.
This approach, however, requires a large DC capacitor to provide sufficient inertia, thus
making it more suitable for large high-voltage DC systems [11]. The combination of the
renewable generation with an electrical energy storage system (ESS) provides the most
flexible approach for establishing grid forming capability as its size can be chosen to
provide adequate support and the support provided does not interfere with the
optimum generation conditions. Under the assumption of sufficient DC side energy
storage, grid forming controls, e.g. virtual synchronous generator (VSG) control [12],
Virtual Synchronous Machine [13] or Synchronverter [14] have been applied to various
different CIG systems. Relevant examples are PV plants [15], rooftop PV systems [16],
wind turbines [17], wind farms [18], vehicle charger stations [19], hybrid PV-diesel
systems [20] and DC Microgrids [21]. These can be integrated with ESSs [22] and
supply droop freqeuncy control and inertia response [23]. Reference [24]
comprehensively reviews the control strategies and implementations of the VSG.
In the case of wind turbines, the placement of an energy storage system, such as a
battery storage on the DC side, requires modification to the wind turbine hardware and
is therefore not an attractive solution for existing wind farms. Alternatively, the ESS
can be placed on the AC side of the wind turbine (i.e. effectively just co-located with the
turbine). In this case, the ESS alone can be controlled to be grid forming and provide
the necessary frequency support during transients similar to the situation where the
ESS is placed on the DC side. For example, reference [25] shows that similar capability
for frequency support can be achieved by placing a VSG controlled ESS on the AC side
of the wind turbine, while leaving the wind turbine configuration unchanged. Although
4
the grid forming and frequency support capability of this approach is similar to the DC
side ESS, it has the major disadvantage that the variable renewable power generation
does not go through the VSG control, so that it has no damping effect on the power
fluctuations. This has an important effect on the overall system frequency variation in
the case of systems with high penetration of variable renewable generation.
1.3 Motivation
In the literature, the ESS is generally assumed to be connected to the DC link of the
generation system, with the generated power and the ESS power sharing the same VSG
controlled converter interfacing to the grid. In this configuration, using VSG control,
the ESS is in the middle (between the generation and grid) and can be considered as a
filter to damp the stochastic power generation with the filtered power being stored in
the ESS. However, the implementation of this approach in the existing CIG, which is
typically operated in grid feeding mode, would require significant changes to the CIG
system configuration which is costly and unnecessary. This work focusses on the
implementation of the grid forming control on AC side ESS placement. Additionally,
most of the previous work has focused on the support function of the grid forming CIG
with DC side energy storage, after being subjected to a large disturbance. However,
besides this transient response, the CIG should also mitigate the effect of the stochastic
variation of the renewable generation on the frequency variance under normal
operation. This is especially important in the situation where the power system is
migrating to 100% CIG penetration [26]. Excessive frequency variation may cause
unwanted tripping of the protection system which might ultimately result in a blackout.
However, none of the previous works have investigated this aspect, nor has there been
any work which shows how AC side ESS grid forming control can be co-ordinated with
the wind turbine control to mitigate this effect. This paper addresses this gap.
1.4 Contribution
5
The main contribution of this work is to propose a novel VSG control approach that
co-ordinates the renewable generator with the ESS placed at the AC side so as to
properly emulate the inertial response of a synchronous machine and that also
effectively damps power variations from the renewable generation. The effectiveness
of this control scheme on reducing system frequency variation in a realistic, high
renewable penetration system is investigated. The proposed control scheme also takes
account of storage limitations and is shown to be effective for small ratios of the
capacity of the ESS versus the capacity of the CIG. The effectiveness of the proposed
approach is thoroughly demonstrated by means of simulation in a real power system,
namely the all-Island Irish transmission system, which is a typical high-wind-
penetration system.
1.5 Organization
The rest of paper is structured as follows: Section II reviews the VSG
implementations and their coordination with the renewable generation. Section III
introduces the proposed control method for the ESS at the AC side of the wind
generation to damp the wind generation considering the rating of the ESS. Section IV
verifies the proposed control via an EMT simulation based on Matlab/Simulink.
Section V presents a system level simulation based on the Irish system and quantifies
the benefits from the proposed control, while section VI draws the conclusions.
2 Virtual Synchronous Generator
The implementation of the VSG control for grid forming CIG has been extensively
described in previous literature and here we only present a brief review necessary for
the understanding of this work. The VSG control is an outer loop applied to the
conventional voltage control of the voltage source converter which provides the
reference for the voltage phase and magnitude. Under the assumption of operation in
6
a largely inductive gird, an active power regulation part determines the reference
voltage phase, and a reactive power regulation part determines the reference voltage
magnitude. A virtual impedance may also be included in the structure as shown in
Figure 1.
Fig. 1. VSG control scheme
In the active power regulation, the swing equation (1) is used for the converter
synchronization and frequency to power droop (2) for the frequency support, where M
is the virtual inertia, D is the damping factor,
!
!
is the droop gain,
"
#
"
is the nominal
frequency,
#
#$$
is the measured frequency from PLL and
#
%&'
is the converter internal
frequency. Note, the PLL here is only used for the droop control rather than the
synchronization, and the measured frequency
#
#$$
follows the grid frequency quickly.
During a transient state, the converter frequency
#
%&'
is delayed in following the grid
frequency change, thereby transiently leading to a larger phase difference δ between
the converter voltage and grid voltage and, thus the output power, P, surges to provide
an emulated inertia response. In steady state, the converter frequency is equal to the
grid frequency and, thus, the power flow remains fixed at the reference. The reference
power, P
*
, is determined by the feed forward power from the renewable generation or
is set to zero if there is no renewable generation and only an ESS connected.
$
%&#
%&'
%' ()
"
*)
!())#
*+&#
%&'
,)
-
.
(
&
#
%&'
/
"
"""""""""""""""""""""""012
"
7
)
!())#
(!
!
0#
"
,"#
#$$
2""""""""""""""""""""""""""""""""""""""""""""""""""032
"
The reactive power regulation is used to determine the emulated electric potential
E of the generator. Voltage to reactive power droop control or automatic voltage
regulation can be applied.
The virtual impedance in the control loop can be considered to connect in series
with the transmission line impedance and could be used to modify the R/X ratio,
increase the system damping to aid stability or decouple the active and reactive power
in a resistive line.
4
)
"
(56-,708
*
*9#
%&'
:
*
2""""""""""""""""""""""""""""""""""""""""""""""""""""0;2
The active power regulation determines the phase of the electric potential, and
reactive power regulation determines the amplitude of the electric potential. After
considering the voltage drop on the virtual impedance, the reference voltage for the
conventional outer voltage, inner current control of the VSC can be obtained as (3).
Here the details of the voltage control of the VSC are omitted as it follows a standard
outer voltage, inner current control scheme implemented in the d-q frame as described
in [27].
The active power P output from the VSG controlled converter consists of the
reference power
)
"
, droop power
)
!())#
, and transient inertia power
0
+&#
%&'
,
$
!+,
!"#
!-
2
, where the droop power and transient inertia power are provided by the ESS.
The ESS can be placed either in the inner DC side (here referred to as VSGi) or the
outer AC side (here referred to as VSGo) of the CIG system. Taking the full converter
(type 4) wind turbine generator (WTG) as an example of the CIG system, the rest of
this section will introduce the characteristics of these two configurations.
8
2.1 Inner energy storage system configuration
In the configuration of the VSGi as shown in Fig. 2, the ESS is placed on the DC side
of the grid-tied converter (G-converter), where the VSG control is implemented. The
machine-side converter (M-converter) operates with the maximum power point
tracking (MPPT) control and is connected to the induction generator (IG), and the ESS-
side converter (E-converter) is used to control the DC voltage. In this topology, the VSG
control of the G-converter forms the voltage
4
)
supplying the power to the grid
4
.
. The
reference power
)
"
in (1) is the renewable generation feed-forward power from the M-
converter, so that the variable renewable generation is an input to the swing equation
emulation and hence undergoes a damping effect. Thus, the whole VSGi system
emulates the behaviour of a conventional SG in terms of power generation and
frequency response. This configuration can damp the power supplied by the G-
converter by means of a bi-directional exchange with the DC side ESS [28].
Fig. 2. VSGi system
2.2 Outer energy storage system configuration
In practice, most wind farms are already connected to the grid operating in grid
feeding mode. For existing systems, the inclusion of the ESS at the DC side would
require an increase in the G-converter capacity, which is costly and unfeasible. A
compromise method is to place the ESS system externally at the AC side of the WTG
system as shown in Fig. 3. In this topology, the original WTG system does not require
9
any change and remains operating in the current source mode feeding the generated
power into the grid.
Fig. 3. VSGo system
The ESS system, although co-located with the WTG, can be regarded as a separate
system and only used to support the frequency. Now with the VSG control applied only
to the ESS, the measured power P in (1) is
)
/00
, reference power
)
"
is set to be zero, and
only the droop power
)
!())#
is an input and is damped by the swing equation of the
VSG controlled E-converter. The grid power injection is simply a summation of the
WTG generated power and ESS compensated power. Importantly, in this case the
generated power is not input to the swing equation and is undamped. Hence, the
dynamics resulting from the stochastic wind generation, act on the grid leading to an
associated frequency variation.
The comparison between the VSGi and VSGo systems in terms their performance
and their associated impacts on the system is presented via a Matlab/Simulink in
Section IV and with a case study for the All-island Irish Transmission System in Section
V.
10
3 Modified virtual synchronous generator control
The undamped power generation from the VSGo system contains significant
oscillations due to the stochastic nature of the wind generation, which results in system
frequency variation. The use of the VSGi configuration can alleviate this frequency
variation due to the inclusion of the damping of the generation. In the VSGo system,
since the ESS is closely linked to the WTG system, if the generated power goes into the
ESS before being injected to the grid via the E-converter with VSG control, then this
power can be damped. Of course, the power flow via the E-converter should be limited
according to the capacity of the converter and storage. This section proposes a control
to achieve such functions.
3.1 Modified VSGo system
Assuming the transmission line is a series inductive impedance
<
.
, and using the
grid voltage
=
.
as a reference at zero rad, the power injected to the grid can be
formulated as (4).
)(=
)
=
.
>7?@
<
.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""0A2
Where
=
)
6@(56-,708
*
*9#
%&'
:
*
2
represents the converter voltage previously
mentioned in (3).
From (4), there are two equations with two parameters (grid voltage and line
impedance), and four variables. If two of these variables are controlled, then the other
two are automatically determined.
In the original WTG system, in grid feeding control, the power is directly controlled
in the G-converter with its value being equal to the generated power. Thus, the
converter voltage amplitude and phase are indirectly changed in order to deliver the
required power. In this scheme, there is another loop needed to damp the power
11
generation from the WTG.
In the ESS system, under VSG control, the voltage amplitude and phase are directly
controlled. The power flow is the consequence of the voltage difference between the
converter and the grid. In this scheme, the phase is determined via the swing equation
(1) with the dynamic response defined by the damping and inertia settings.
According to the above analysis, if we want to damp the wind power generation
before injecting to the grid, the ESS has to determine the voltage at the point of
common coupling (PCC) of the whole VSGo system in Fig. 3 and regulate the power
flow between this point and the grid. On the other hand, in steady state, the power flow
from the PCC to the grid should equal the WTG generation and ESS should not supply
any power. To achieve this, the reference power in (1) can be changed to be the
measured power being injected to the PCC from the WTG.
Fig. 3 can be used to illustrate the power generation in this configuration. Defining
the generated power from the WTG is
)
1.
, the power from the ESS system is
)
/00
, and
power injected into the grid is
)
.
. Initially assume that the grid is working at nominal
frequency and no droop power is needed. At this time, the power being injected to the
grid is the generated power and the ESS power is zero, while the reference power in (1)
is the measured WTG power, i.e. (5).
)
1.
()
"
()
.2
()
1.3
B"""")
/00
(C"""""""""""""""""""""""""""""""""""""""0D2
Subsequently, considering that the generated power from the WTG changes by
E)
,
due to the damping and inertia (1), the phase from the PCC voltage to the grid cannot
change instantly, thus the grid power injection remains the same. The change in
generated power goes into and is stored by the ESS.
)
1.
()
1.3
*E)B"")
"
()
.
()
1.3
B""")
/00
(,E)""""""""""""""""""""""""0F2
After the VSG control detects the generation change and adjusts the reference
12
power, the phase changes smoothly as a consequence of the unbalanced power (
)
"
G
)
.
) in the swing equation and leads to the grid power injection changing to the
generation value; this changed grid power injection comes from the ESS as indicated
by (7). Here
H0'2
is the transfer function from the reference power to the real power
output, which is related to the virtual inertia, damping and impedance, detailed in [29]:
)
1.
()
"
()
1.3
*E)B"")
.
()
1.3
*E)IH0'2B""")
/00
(,E)*E)IH0'2"""""""0J2
When the output power flow equals the generated power, the ESS returns to a zero
input, and the system stabilizes to a new steady operation point as presented in (8):
)
1.
()
"
()
.
()
1.3
*E)B"""""")
/00
(C"""""""""""""""""""""""""""""""""0K2
In this process, the generated power is damped by the VSG controlled E-converter
of the ESS. Since the swing equation becomes the same as that used in the VSGi
operation, they would have the same response as viewed from the PCC point, even
though their internal control structures are different
3.2 Excessive Power Regulation
In the proposed VSGo configuration, the change in power
E
)
initially all goes into
the ESS. Considering that in the worst case the maximum power change could be equal
to the rating of the WTG, then the rating of the E-converter should be the same as the
WTG in order to safely damp all the generated power. However, the principal function
of the ESS is to provide frequency response, and for this function, the rating may not
need to be so large. In this case, the maximum change in power cannot fully pass
through the E-converter to be injected into the storage but can only partially do so. In
this case, the system control should allow the maximum power which respects the E-
converter rating to be injected into the storage, thus damping the generation as much
as possible.
Defining
)
4
as the rating of the E-converter, if the changed power
E
)
detected in
(9) is greater than the ESS rating
)
4
, then the excess power
)
/5
(the difference between
13
the power change and the rating of the E-converter) should be injected into the grid
directly and not passed via the ESS:
E)()
1.
,)""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""0L2
)
/5
(ME)M,)
4
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""01C2
The PCC voltage is actively controlled and the power flowing into the grid is the
consequence of the voltage difference in (4), where the converter voltage is determined
by the VSG control in (1~3). In this case, in order to instantly deliver the excess power
)
/5
, the phase
@
has to change instantly. A PI controller (11) is used for this purpose as
shown in Fig. 4, where
!
#
is used to link
)
/5
to
@
, and
!
6
is used to ensure the power is
limited in the ESS as discussed in the next section.
H
78
0>2(!
#
*!
6
>"""""""""""""""""""""""""""""""""""""""""""""""""""0112
3.3 PI Parameter Design
The VSG control can only directly regulate the phase
-
of the electric potential,
while the converter output voltage phase
@
is shifted by the virtual impedance.
According to [29], the transfer function of the power change
E
)
.
to voltage phase
change
E-
, considering the influence of the virtual impedance, can be expressed as:
N
9
0>2(EO
E-(;5
3
3P
:
>
;
*P
;
>*P
<
0Q
;
*<
;
2R0Q*>S2
;
*<
;
T"""""""""""""""""""""0132
"
where
"
P
:
(
U
.
3
S
S
.
0
Q>7?
-
3
*
<VW>
-
3
2
,
5
3
S<
S
.
B
P
;
(3U
.3
S8
.
0Q>7?-
3
*<VW>-
3
2,
35
3
S<8
.
,U
.3
S
*
>7?-
3
0Q
;
,<
;
2*35
3
S
*
<>7?-
3
0Q>7?-
3
,<VW>-
3
2
B
P
<
(
U
.
3
0
Q
;
*
<
;
2
0
Q>7?
-
3
*
<VW>
-
3
2
,
3
U
.
3
Q
*
Q
;
>7?
-
3
*
3
5
3
Q
*
<>7?
-
3
0
Q>7?
-
3
,
<VW>
-
3
2
XYZ"
where
5
3
,
"=
.3
,
-
3
are the initial equilibrium point values, and
Q
(Q
*
*Q
.
,
<
(<
*
*<
.
,
S
(
S
*
*S
.
. The poles of
N
96=/
0>2
are at
,
>
9
[9#
.
, for which resonance occurs around the
14
grid synchronous frequency
#
.
, where the phase changes from 0
º
to -180
º
.
The transfer function of the outer voltage inner current controls can be expressed
as:
H
%&?
0>2( -
-
(/@
(!
#*
>*!
6*
\
@
'
6
>
<
*\
@
>
;
*!
#*
>*!
6*
"""""""""""""""""""""""""""""01;2
Where
'
6
is the time constant of the closed-loop current controller,
\
@
is LC filter
capacitance,
!
#*
]!
6*
are the PI controller parameters of the voltage controller. The
time constant of
H
%&?
0>2
is normally less than 10 ms, i.e. around 100 Hz, the phase of
H
%&?
0>2
change from 0
º
to -180
º
and the gain becomes negative from 0 dB.
The open-loop transfer function of the excess power regulation is:
H
AB
0>2(0!
#
>*!
6
>2H
%&?
0>2N
9
0>2""""""""""""""""""""""""""""""""""""01A2
When the frequency approaches
#
.
, the phase of
H
AB
0>2
turns to -180
º
and even
lower due to
H
%&?
0>2
and
N
9
0>2
. The PI controller of the excess power regulation has a
corner frequency at
C
$
C
%
, where the phase turns from -90
º
to 0
º
and the slope turns from
-20 dB/s to 0 dB/s. In order to ensure a satisfactory phase margin, the corner
frequency of the PI controller should be less than
#
.
. In this situation, at the lower
frequency the system is mainly dominated by the PI controller, thus,
H
%&?
0>2
is
negligible and
N
96=/
0>2
acts as a pure gain:
N
9D3
(;
35
3
P
<
0Q
;
*<
;
2
;
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""01D2
"
The virtual impedance influences the value of
N
9D3
, the rationale for which has been
detailed in [
3
0]. The open-loop transfer function of the excess power regulation can be
simplified to a second order system with crossover frequency:
H
AB
0
>
2
(
^
!
#
>
*
!
6
>
_
N
96=/D3
""""""""""""""""""""""""""""""""""""""""""""01F2
15
#
EF
(
`
!
6
;
N
9
;
1
,
!
#
;
N
9
;
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
0
1J
2
In order to ensure the slope at crossover is greater than -20 dB/decade,
#
EF
should
be less than the corner frequency
C
$
C
%
. Therefore, the condition for a stable operation of
the excess power regulation can be summarized as:
`
!
6
;
1
,
!
#
;
N
9
;
a
!
6
!
#
a
#
.
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
"
0
1K
2
Based on (16), the closed-loop transfer function of the excess power regulation is:
H
ABDG
0>2(" N
9
!
#
>*N
9
!
6
0N
9
!
#
*12>*N
9
!
6
"""""""""""""""""""""""""""""""""""""""""""""""""""01L2
The time constant of (19) is:
'
6
(A0N
9
!
#
*12
N
9
!
6
"""""""""""""""""""""""""""""""""""""""""""""""""""03C2
The time constant of the excessive power regulation should be as short as possible
in order to quickly limit the ESS output.
When the power change is within the ESS limit, solely the VSG control is activated,
as shown in Fig. 4, the transfer function
b0c2"
of the VSG control used in (7) is:
H
%&'
0
>
2
(
^
1
𝑀𝑠
+
𝐷
∙
1
𝑠
_
H
%&?
0
>
2
N
96=/
0
>
2
"
""""""""""""""""""""""""""""""""""0312
for which the stability was analysed in [19]. The VSG control scheme has to be modified
to include the excess power regulation part, as shown in Fig. 4, where the blue part is
the VSG control used to damp the unsaturated power generation, while the red part is
the excess power regulation used to deliver the saturated power to the grid directly.
This modified VSG control scheme is to replace the conventional VSG control scheme
16
in the active power regulation of Fig. 1 and control the converter phase angle δ.
Fig. 4. Transfer function of the modified VSGo system.
4 EMT Simulation Validation
The proposed WTG configuration and its control are simulated in Matlab/Simulink
using the EMT model of the converters. The WTG connects to the grid through a line
impedance, where the wind turbine and battery in this simulation are assumed to be
an ideal DC source and the grid is assumed to be an ideal AC source. The converter
parameters and VSG parameters are given in Table I. There are two tests presented
here. The first test compares the proposed WTG configuration with the other
configurations, i.e., VSGo and VSGi, regardless of the rating of the ESS. The second test
considers the rating of the ESS and verifies the operation of the proposed excessive
power regulation.
TABLE I. VSG settings
Parameter
Value
Parameter
Value
Sampling time
1e-7 s
PWM rate
1350
Rated Voltage 𝒗𝒈
(
8165 V
Filter inductance
0.1 H
Reference voltage
𝑽∗
8165 V
Filter resistance
0.12 Ω
Reference angular frequency 𝝎∗
2π*50 Hz
Filter capacitance
13 µF
VSG Inertia 𝐌
2.6 kW/(rad·s-2)
Line inductance
0.1 H
damping/droop 𝑲𝒅/𝑫
159 kW/(rad·s-1)
Line resistance
0.01 Ω
Excess power controller P/I
3E-7 /5E-5
Virtual inductance
0 H
Current controller P/I
222/326
Virtual resistance
15 Ω
VSG and WTG PLL P/I
0.022/0.392
Voltage controller P/I
0.008/ 1.87
4.1 Test 1: wind turbine generator configurations
−
𝑃
𝑤𝑔
+
𝐺𝑃𝐼(𝑠)
−𝑃
𝑚
𝑃
𝑚
+
+
VSG
Control
Excessive Power
Regulation
+
−
DM +s
1
)(sH L
)(sGVSC
d
PD
s
1
17
The aim of this test is to verify that the proposed control approach can make the
outer ESS WTG combination have the same effect as the inner ESS WTG on both the
power generation and frequency support. The VSGi and VSGo are used for comparison.
To produce a fair comparison, the parameters are set identically for all configurations.
The rating of the ESS is not limited in this test, thus, the power limit
)
4
in the proposed
control (Fig. 4) is set to be infinite. The tested system experiences a wind power
generation increase from 0 to 500 kW at 0.5 s and a grid frequency reduction from 50
Hz to 49.9 Hz with a 10 Hz/s ramp at 2 s. Fig. 5 shows the resulting grid power injection
at the
4
.
in Fig. 2 and Fig. 3.
It can be seen from Fig. 5 that the power output in response to the grid frequency
change is the same in all configurations (2~3.5 s), which indicates that similar
frequency support can be provided regardless of the configuration of the WTG with
ESS. This is as expected because the frequency variation presented to the swing
equation (1) and droop (2) leads to power variation regardless of the ESS placement.
However, the VSGi presents damping on the power generation, as its generated power
goes through the swing equation (1), i.e.
)
"
()
1
. Without this action, i.e.
)
"
(C
, the
generated power is injected into the grid instantly with a step change as presented in
the VSGo case. In this context, the proposed method can make the VSGo behave like
VSGi by forcing
)
"
()
1
and resulting in the generated wind power flowing through the
E-converter to the grid with the same damping. This verifies that the proposed method
can add damping on the generation for the external ESS configuration, and the
generated power flows accordingly (5~8) as expected.
18
Fig. 5. Test 1: Grid power injection from different WTG configurations.
4.2 Test 2: Excess Power Regulation
This test aims to validate the excess power regulation function of the proposed
method. In this test, the active power via the E-converter is limited at 200 kW,
corresponding to
)
4
(3CC"de
in Fig. 5. At 0.5 s, the generated power step increases
to 500 kW from 0 kW. Fig. 6 shows the power from the E-converter
)
/00
, from the wind
generator or G-converter
)
1.
, and to grid
)
.
.
Fig. 6. Test 2: excess power regulation.
Figure 6 shows that the grid power injection
)
.
almost step changes (50 ms settling
time as the designed reaction of the controller) to 300 kW at the instant of the
generated power step change
)
1.
at 0.5 s, so that the power to the ESS
)
/00
is
00.5 11.5 22.5 33.5
0
500
1000
Time (s)
Pg (kW)
VSGo
VSGi
Proposed
00.5 11.5 2
-200
0
200
400
Time (s)
Active Power (kW)
Pess Pwg Pg
19
suppressed within its limit. The rest of the power, 200 kW, still flows into the ESS at
first and is damped by the E-converter of the ESS. This verifies the effectiveness of the
proposed excess power regulation in respect with the rating limits for the operation of
the ESS.
4.3 Test 3: Parameter Design
This test aims to verify the parameter design of the excessive power regulation.
Based on the VSG settings listed in Table I, there are three cases with different PI
parameters for the excessive power regulation as given in Table II, where:
•
Case 1 is the base case which strictly fulfils the rules in Section 3.3 that the
stability constrains are satisfied and the controller time constant is minimized.
•
Case 2 is a comparative case which decreases
!
6
for a larger controller time
constant but the stability constrains are still satisfied.
•
Case 3 is another comparative case which breaks the stability constraints as the
crossover frequency is greater than
#
.
.
TABLE II. Excessive power regulation PI settings for Test 2
𝑲
𝒑
𝑲
𝒊
𝝎
𝐜𝐨
(rad/s)
𝒕
𝒊
(ms)
Case 1
3E-7
5E-5
206
55
Case 2
3E-7
5E-6
20.6
550
Case 3
3E-7
5E-4
2060
5.5
The system experiences the same changes as used in Test 2. Figure 7 shows the bode
plot of the cases and Fig. 8 shows the results of the ESS output power in the time
domain simulation. When the stability constraints are satisfied, the crossover
frequency can be roughly predicted as shown in Fig. 7, and the system is stable as
proved in Fig. 8. The crossover frequency is closer to
#
.
, the stability margin is lowered.
20
However, since the time constant of the Case 2 is greater than that of the Case 1, Case
2 shows a higher peak power at the instant of the power change while Case 1 can better
limit the ESS power output as shown in Fig. 8. On the other hand, when the stability
constraints are not satisfied, i.e.
#
EF
f#
H
as in Case 3, the phase sharply decreases
below -180º at the synchronous frequency while the gain is still positive as shown in
Fig. 7, thus the system is unstable as shown in Fig. 8. Note, in this case, the crossover
frequency cannot be precisely predicted since the effect of the synchronous resonance
is significant, and the transfer function analysis must use (14) but not (16).
Fig. 7. Bode plot of these cases in Test 3.
-150
-100
-50
0
50
100
Magnitude (dB)
Frequency (rad/s)
100101102103104105
-360
-270
-180
-90
0
Phase (deg)
Case 1 Case 2 Case 3
(21, -42)
(317, -119)
(601, -215)
21
Fig. 8. Test 3: Parameter design.
5 Case Study
The Matlab/Simulink simulation in Section 4 is based on a two-bus system, which
tests the CIG connected to an infinite bus, but this does not show the effect of the
damped generation on the system frequency variations. In this section, we use the Irish
transmission system as an example to demonstrate these effects, and furthermore to
quantify the minimum total rating of the storage in the case of 100% wind penetration,
in order to restrict the frequency variations within 50±0.2 Hz according to the Irish
grid code requirement.
The model of the Irish transmission system consists of 1479 buses, 1851
transmission lines and transformers, and 245 loads. The dynamic models include 21
conventional synchronous power plants and 176 wind power plants. Reference [26]
shows that the replacement of the SG in the system by VSG-controlled WTG with
identical settings could achieve 100% wind penetration with the same stability level.
This paper focuses on the frequency variations resulting from the VSG controlled
WTG and storage combination in a 100% non-synchronous generation case, thus, all
of the SGs are replaced by the WTGs combined with ESS as given in [26]. Note, in the
case study, the VSG uses AVR for the voltage stability. The overall system loading is
2.36 GW. The wind input is set to be stochastic using the Weibull distribution model
for all the WTGs and the load is also set to be stochastic using the Ornastein-
0.5 0.6 0.7
-500
0
500
1000
1500
Time (s)
ESS Active Power (kW)
Case 1
Case 2
Case 3
22
Uhlenbeck’s process [31] with ±5% of variation to their original values.
A Monte Carlo analysis is used where 100 simulations are run for each case in
DOME, a Python-based power system software [32]. The grid frequency is measured
by the centre of inertia (COI). Two cases are presented in the paper, the first one verifies
the effectiveness of the proposed outer ESS WTG operation strategy, compared with
the VSGo and VSGi results; the second one analyses the effect of the ESS rating on the
frequency variations and, particularly, quantifies the minimum rating of the ESS
required to damp the generation at this loading.
5.1 Case 1: wind turbine generator configurations
In this case, the paper compares the frequency variations from the stochastic wind
and loads in the Irish system under different WTG configurations, i.e., VSGo, VSGi and
the proposed method. The rating of the ESS is not limited in this case. Fig. 9 shows that
the use of VSGo in the Irish system leads to a significant frequency variance of ±0.26
Hz, while the application of the VSGi can reduce this variance by 50% to ±0.13 Hz. Fig.
9(c) verifies the effectiveness of the proposed control to reduce frequency fluctuations
compared to the conventional VSGo scheme. The dynamic behaviour of the proposed
control is comparable with that of the conventional VSGi scheme. It is noticed that for
the Irish system, the normal system frequency variance should be within ±0.2 Hz.
Clearly, the use of the VSGo with conventional control does not satisfy this but the
VGSo with the proposed control does.
23
Fig. 9. Case 1 result: WTG configurations effects on the frequency variations.
In order to illustrate the origin of the different frequency variance in these
configurations, Fig. 10 presents the grid power injection from one of the WTG systems
in one of the scenarios. It can be seen that, compared with the VSGo, the power
injection from the VSGi is smoother. The proposed method achieves the same power
response as the VSGi. The variation of power injected to the grid results in the
frequency variance. This is the reason that the VSGo shows higher system frequency
variance than the VSGi and the proposed method.
𝑓
! (Hz)
𝑓
! (Hz)
Average
Variance
Average
Variance
𝑓
! (Hz)
Average
Variance
24
Fig. 10. Case 1 result: Grid power injection from WTG systems.
5.2 Case 2: limited energy storage system rating
In the previous results, the proposed VSG control has been shown to damp the wind
generation from the co-located ESS, and limit the ESS power flow to avoid overload of
the ESS. This case aims to investigate the effect of the ESS rating on the frequency
variance. To investigate this the rating of the ESS is gradually reduced from 100% of
the WTG initial power to 0% and the results of the frequency variance are recorded in
Fig. 11.
Figure 11 shows that the increase of the ESS rating from 0% provides a considerable
improvement in the frequency variance initially, but this improvement becomes
saturated with further increase in the ESS rating with no further improvement is
achieved beyond an ESS rating greater approximately 20%. In other words, as long as
the ESS rating is greater than 21% of the WTG power generation, then it is possible to
damp all of the generated power. This is because only the change in generated power
flows through the ESS, rather than the initial power which is delivered to the grid
directly.
25
Fig. 11. Case 2 result: ESS rating vs. frequency variance.
To meet the requirement of the grid code on the ±0.2 Hz frequency variance, the
rating of the ESS should be at least 5% of the WTG initial power. On the other hand, it
should be noticed that the frequency variance in the no ESS situation (1.21 Hz) is higher
than that in the VSGo situation presented in case 1 (0.26 Hz). This is because, in the
case of the VGSo, although the generated power is not damped, the resulting frequency
variance is still reacted to by the ESS and this helps alleviate this variance. However,
due to the inertia and damping, the VSG time constant (2 s) is greater than the
frequency of the wind oscillation (0.5 s), so that the ability of this frequency support is
limited. Most of the frequency variance reduction should still rely on the damping of
the generation directly, which is achieved by the proposed method.
Figure 12 presents the grid power injection of one of the WTG systems in one of the
scenarios for various different ESS ratings of 4.7%, 13.4% and 100%. The increase in
the ESS rating smooths the grid power injection. Notably, the 13.4% ESS rating
scenario has a very similar performance to the power generation with the 100%
scenario.
···
100
26
Fig. 12. Case 2 result: Grid power injection from WTG systems.
In order to further explain such a phenomenon that a small quantity of the ESS can
adequately damp the wind generation dynamics, the probability density of the wind
generation at one of the WTG systems in all the 100 scenarios has been plotted in Fig.
11, where the x-axis is the generated power variation expressed as a percentage of its
mean value and the y-axis is the probability of occurrence of each of the variations. This
wind power variation is computed from the power deviation from its average using the
case with no ESS. Nearly 90% of the wind generation variation occurs within a ±20%
band of the mean value, thus a corresponding value of the ESS rating (22.6% in Fig. 11)
can almost damp all of the generation. The highest probability of the wind power
variation is around ±5%, which is the reason for the curve of Fig. 9 to exhibit a sharp
change in the frequency variance level when the ESS comprises around 5% of the WTG
initial power.
Fig. 13. Probability of the wind power variation in percentage.
27
6 Conclusions
This paper proposes a modified virtual-synchronous-generator control method for
the outer energy storage system co-located with wind generators. The proposed
coordinated control effectively damps the power fluctuations of the wind turbines and
properly takes into account the limited capacity of the energy storage system.
Importantly, the proposed control method only involves the energy storage system and
does not require any modification in the controllers of the wind power plant. Yet, it
achieves the same performance as the system where the storage is connected internally
to the DC side of the converters of the wind turbines. The case study of the all-island
Irish transmission system serves to prove that the proposed method can effectively
mitigate the frequency variations caused by the stochastic renewable generation and
only a relatively small capacity of the energy storage system is required to make the
proposed control effective. The results would indicate that the storage capacity needs
to be about the size of the variance of the stochastic fluctuations of the power generated
by the wind power plants, e.g., in Irish system, if the power rating of the ESS is over 5%
of the WTG initial power, the grid code requirements relating to frequency variance
can be satisfied.
Acknowledgements
This work is partly funded by the Science Foundation Ireland (SFI) under grant
numbers SFI/15/SPP/E3125 and SFI/15/IA/3074,and by the European
Commission, under the project EdgeFLEx, grant no. 883710.
.
References
[1] J. Đaković, M. Krpan, P. Ilak, T. Baškarad, and I. Kuzle, “Impact of wind capacity share, allocation
of inertia and grid configuration on transient RoCoF: The case of the Croatian power system,”
International Journal of Electrical Power & Energy Systems, Vol. 121, Oct. 2020, pp.1-8.
[2] F. Milano, F. Dörfler, G. Hug, D. J. Hill and G. Verbič, "Foundations and Challenges of Low-Inertia
Systems (Invited Paper)," 2018 Power Systems Computation Conference (PSCC), Dublin, Ireland, 11-
15 June, 2018, pp. 1-25.
[3] J. Liu, Z. Yang, J. Yu, J. Huang, and W. Li, “Coordinated control parameter setting of DFIG wind
farms with virtual inertia control,” International Journal of Electrical Power & Energy Systems, vol.
122, pp. 1-11, November 2020.
28
[4] H. Wang, Z. Chen, and Q. Jiang, “Optimal control method for wind farm to support temporary
primary frequency control with minimised wind energy cost,” IET Renewable. Power Generation,
vol. 9, no. 4, pp. 350–359, May 2015.
[5] Y. Wang, J. Meng, X. Zhang and L. Xu, "Control of PMSG-Based Wind Turbines for System Inertial
Response and Power Oscillation Damping," IEEE Transactions on Sustainable Energy, vol. 6, no. 2,
pp. 565-574, April 2015.
[6] D. Ochoa and S. Martinez, "Fast-Frequency Response Provided by DFIG-Wind Turbines and its
Impact on the Grid," IEEE Transactions on Power Systems, vol. 32, no. 5, pp. 4002-4011, Sept. 2017.
[7] I. Cvetkovic, D. Boroyevich, R. Burgos, Chi Li, M. Jaksic and P. Mattavelli, "Modeling of a virtual
synchronous machine-based grid-interface converter for renewable energy systems integration,"
2014 IEEE 15th Workshop on Control and Modeling for Power Electronics (COMPEL), Santander,
Spain, 22-25 June, 2014, pp. 1-7.
[8] L. Huang, H. Xin, Z. Wang, K. Wu, H. Wang, J. Hu, et al., "A Virtual Synchronous Control for
Voltage-Source Converters Utilizing Dynamics of DC-Link Capacitor to Realize Self-
Synchronization," IEEE Journal of Emerging and Selected Topics in Power Electronics, vol. 5, no. 4,
pp. 1565-1577, Dec. 2017.
[9] J. Fang, H. Li, Y. Tang and F. Blaabjerg, "Distributed Power System Virtual Inertia Implemented
by Grid-Connected Power Converters," IEEE Transactions on Power Electronics, vol. 33, no. 10, pp.
8488-8499, Oct. 2018.
[10] J. Fang, H. Li, Y. Tang and F. Blaabjerg, "Distributed Power System Virtual Inertia Implemented
by Grid-Connected Power Converters," IEEE Transactions on Power Electronics, vol. 33, no. 10, pp.
8488-8499, Oct. 2018.
[11] Y. Cao, W. Wang, Y. Li, Y. Tan, C. Chen, L. He, et al., "A Virtual Synchronous Generator Control
Strategy for VSC-MTDC Systems," IEEE Transactions on Energy Conversion, vol. 33, no. 2, pp. 750-
761, June 2018.
[12] T. Shintai, Y. Miura and T. Ise, "Oscillation Damping of a Distributed Generator Using a Virtual
Synchronous Generator," IEEE Transactions on Power Delivery, vol. 29, no. 2, pp. 668-676, April
2014.
[13] S. D’Arco, J. A. Suul, and O. B. Fosso, “A Virtual Synchronous Machine implementation for
distributed control of power converters in SmartGrids,” Electric Power Systems Research, vol. 122,
pp. 180-197, May 2015.
[14] Q. Zhong and G. Weiss, "Synchronverters: Inverters That Mimic Synchronous Generators," IEEE
Transactions on Industrial Electronics, vol. 58, no. 4, pp. 1259-1267, April 2011.
[15] W. Ming and Q. Zhong, "Synchronverter-based transformerless PV inverters," IECON 2014 -
40th Annual Conference of the IEEE Industrial Electronics Society, Dallas, TX, 29 Oct.-1 Nov.,
2014, pp. 4396-4401.
[16] S. Mishra, D. Pullaguram, S. A. Buragappu, and D. Ramasubramanian, “Single-phase
synchronverter for a gridconnected roof top photovoltaic system,” IET Renewable Power Generation,
vol. 10, no. 8, pp. 1187-1194, May 2016
[17] Q. Zhong, "Virtual Synchronous Machines: A unified interface for grid integration," IEEE Power
Electronics Magazine, vol. 3, no. 4, pp. 18-27, Dec. 2016.
[18] Q. C. Zhong, Z. Ma, W. L. Ming, and G. C. Konstantopoulos, “Grid-friendly wind power systems
based on the synchronverter technology,” Energy Conversion and Management, vol. 89, pp.719-726,
January 2015.
[19] J. A. Suul, S. D'Arco and G. Guidi, "Virtual Synchronous Machine-Based Control of a Single-
Phase Bi-Directional Battery Charger for Providing Vehicle-to-Grid Services," IEEE Transactions on
Industry Applications, vol. 52, no. 4, pp. 3234-3244, July-Aug. 2016.
[20] A. Belila, Y. Amirat, M. Benbouzid, E. M. Berkouk, and G. Yao, “Virtual synchronous generators
for voltage synchronization of a hybrid PV-diesel power system,” International Journal of Electrical
Power & Energy Systems, vol.117, pp.1-14, May 2020.
[21] D. Chen, Y. Xu and A. Q. Huang, "Integration of DC Microgrids as Virtual Synchronous Machines
Into the AC Grid," IEEE Transactions on Industrial Electronics, vol. 64, no. 9, pp. 7455-7466, Sept.
2017.
29
[22] J. Chen, M. Liu and T. O'Donnell, "Replacement of Synchronous Generator by Virtual
Synchronous Generator in the Conventional Power System," 2019 IEEE Power & Energy Society
General Meeting (PESGM), Atlanta, GA, USA, 4-8 Aug., 2019, pp. 1-5.
[23] K. M. Cheema, A. H. Milyani, A. M. EI-Sherbeeny, and M. A. EI-Meligy, “Modification in active
power-frequency loop of virtual synchronous generator to improve the transient stability,”
International Journal of Electrical Power & Energy Systems, vol. 128, pp. 1-9, June 2021.
[24] K. M. Cheema, “A comprehensive review of virtual synchronous generator,” International
Journal of Electrical Power and Energy Systems, vol. 120, pp.1-10, September 2020.
[25] J. Chen, M. Liu, F. Milano and T. O’Donnell, "Placement of Virtual Synchronous Generator
Controlled Electric Storage combined with Renewable Generation," 2019 IEEE Milan PowerTech,
Milan, Italy, 23-27 June 2019, pp. 1-6.
[26] J. Chen, M. Liu, F. Milano, and T. O’Donnell, “100% Converter-Interfaced generation using
virtual synchronous generator control: A case study based on the irish system,” Electric Power
Systems Research, vol. 187, pp. 1-10, Octorber 2020.
[27] A. Yazdani and R. Iravani, “Voltage-Sourced Converters in Power Systems: Modeling, Control,
and Applications”, John Wiley & Sons, Canada, ISBN 978-0-470-52156-4, 2010.
[28] Y. Yang, J. Xu, C. Li, T. Dragicevic, and F. Blaabjerg, “Stability enhancement of DC power
systems by VSM control strategy,” International Journal of Electrical Power & Energy Systems,
vol. 126, p. 106569, 2021.
[29] J. Chen and T. O'Donnell, "Parameter Constraints for Virtual Synchronous Generator
Considering Stability," IEEE Transactions on Power Systems, vol. 34, no. 3, pp. 2479-2481, May
2019.
[30] J. Chen and T. O’Donnell, “Analysis of virtual synchronous generator control and its response
based on transfer functions,” IET Power Electronics, vol. 12, no. 11, pp. 2965–2977, 2019.
[31] F. Milano and R. Zárate-Miñano, "A Systematic Method to Model Power Systems as Stochastic
Differential Algebraic Equations," IEEE Transactions on Power Systems, vol. 28, no. 4, pp. 4537-
4544, Nov. 2013.
[32] F. Milano, "A python-based software tool for power system analysis," 2013 IEEE Power &
Energy Society General Meeting, Vancouver, BC, 21-25 July, 2013, pp. 1-5.
